Re: [PATCH 4/4] iommu/amd: Introduce boot option ivmd=seg:bus:dev.fun,start,size,flags

[Suthikulpanit, Suravee]

On 4/22/2026 5:56 PM, Yu Zhang wrote:
> On Mon, Apr 20, 2026 at 05:00:33PM +0000, Suravee Suthikulpanit wrote:
> > IVRS table contains IVMD blocks, which allow firmware to specify memory
> > usage requirements to communicate to system software based on its needs
> > or on hardware characteristics. Each IVMD entry may be per-device, range
> > of devices.
> >
> > Some BIOS specify incorrect or missing IVMD entry. Introduce a new ivmd
> > boot option to allow user to specify up-to 4 per-device IVMD entries at
> > boot time. The entries are stored during driver initialization, and will
> > be added to the per-segment ivmd_entry_map so that they can be included
> > during struct iommu_ops.get_resv_regions.
> >
> > Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@xxxxxxx>
> > ---
> >   .../admin-guide/kernel-parameters.txt         |  16 +++
> >   drivers/iommu/amd/amd_iommu_types.h           |  14 +++
> >   drivers/iommu/amd/init.c                      | 104 +++++++++++++++++-
> >   3 files changed, 133 insertions(+), 1 deletion(-)
> >
> > diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
> > index 03a550630644..e680a258008f 100644
> > --- a/Documentation/admin-guide/kernel-parameters.txt
> > +++ b/Documentation/admin-guide/kernel-parameters.txt
> > @@ -2897,6 +2897,22 @@ Kernel parameters
> >   			  PCI device ID 00:14.5, write the parameter as:
> >   				ivrs_acpihid[0001:00:14.5]=AMD0020:0
> >   
> > +	ivmd		[HW,X86-64]
> > +			Supplement IVMD unity mapping or exclusion ranges from
> > +			the kernel command line (in addition to the IVRS ACPI
> > +			table). May be specified multiple times.
> > +
> > +			Form:
> > +				ivmd=segment:bus:dev.fn,start,length,flags
> > +
> > +			start and length are byte counts (decimal or 0x hex).
> > +			flags is the IVMD flags byte (UNITY, IR, IW, EXCL bits
> > +			per the AMD IOMMU specification). Use segment 0 for
> > +			PCI segment zero.
> > +
> > +			Example:
> > +				ivmd=0000:00:01.0,0xe0000000,0x100000,0xb
> > +
> >   	js=		[HW,JOY] Analog joystick
> >   			See Documentation/input/joydev/joystick.rst.
> >   
> [...]
>
> > diff --git a/drivers/iommu/amd/init.c b/drivers/iommu/amd/init.c
> > index 4b62bb89a12c..7d60143d4711 100644
> > --- a/drivers/iommu/amd/init.c
> > +++ b/drivers/iommu/amd/init.c
> > @@ -208,10 +208,12 @@ enum iommu_init_state {
> >   static struct devid_map __initdata early_ioapic_map[EARLY_MAP_SIZE];
> >   static struct devid_map __initdata early_hpet_map[EARLY_MAP_SIZE];
> >   static struct acpihid_map_entry __initdata early_acpihid_map[EARLY_MAP_SIZE];
> > +static struct ivmd_cmdline early_ivmd_cmdline_map[EARLY_MAP_SIZE] __initdata;
> >   
> >   static int __initdata early_ioapic_map_size;
> >   static int __initdata early_hpet_map_size;
> >   static int __initdata early_acpihid_map_size;
> > +static int early_ivmd_cmdline_map_size __initdata;
> >   
> >   static bool __initdata cmdline_maps;
> >   
> > @@ -2653,6 +2655,47 @@ static int __init init_ivmd_map_range(struct ivmd_header *m,
> >   	return 0;
> >   }
> >   
> > +static int __init apply_ivmd_cmdline_entries(struct acpi_table_header *ivrs_base)
> > +{
> > +	int i;
> > +	struct ivmd_entry *e;
> > +	struct amd_iommu_pci_seg *pci_seg;
> > +
> > +	for (i = 0; i < early_ivmd_cmdline_map_size; i++) {
> > +		struct ivmd_cmdline *cmd = &early_ivmd_cmdline_map[i];
> > +
> > +		pci_seg = get_pci_segment(cmd->pci_seg, ivrs_base);
> > +		if (!pci_seg) {
> > +			pr_err("ivmd: PCI segment %#x unavailable\n",
> > +			       cmd->pci_seg);
> > +			continue;
> > +		}
> > +
> > +		if (cmd->devid > pci_seg->last_bdf) {
> > +			pr_err("%s: requestor %#x:%#02x:%#02x.%#02x exceeds segment %#x last BDF %#x\n",
> > +			       __func__, cmd->pci_seg, PCI_BUS_NUM(cmd->devid),
> > +			       PCI_SLOT(cmd->devid), PCI_FUNC(cmd->devid),
> > +			       pci_seg->id, pci_seg->last_bdf);
> > +			continue;
> > +		}
> > +
> > +		e = kzalloc(sizeof(*e), GFP_KERNEL);
> > +		if (!e) {
> > +			kfree(cmd);
>
> cmd points to the early_ivmd_cmdline_map, shall not be kfreed.

Correct, it was from my older implementation. Thanks for catching. I'll 
fix it in v2.

> > +			return -ENOMEM;
> > +		}
> [...]
>
> > +static int __init parse_ivmd(char *str)
> > +{
> > +	u32 seg, bus, dev, fn;
> > +	unsigned long long range_start, range_len;
> > +	unsigned int flags;
> > +	struct ivmd_cmdline *cmd;
> > +
> > +	if (!str)
> > +		goto invalid;
> > +
> > +	if (sscanf(str, "%x:%x:%x.%x,%llu,%llu,%x",
> > +		   &seg, &bus, &dev, &fn, &range_start,
> > +		   &range_len, &flags) != 7)
>
> sscanf with %llu cannot parse values with 0x prefix, which the doc above claims
> to accept.

I will fix the implementation / documentation to only allow hexadecimal 
only.

Thanks,
Suravee

> > +		goto invalid;
> > +
>
> B.R.
> Yu